Three Buses

There’s a well-known British gripe about the bus service—you wait forever, then three come at once.

I’m using that to go a little further down the automation road. The UK Labour Party conference took place last week, and Jeremy Corbyn, who since the June 2017 election fancies his chances, spoke to his audience about plans for a robot tax.

The reaction from the manufacturing sector and the Tory press was swift—the word Luddite was very much in evidence.  The left, of course, was quick to support the idea. When we compare the two articles (using the Private Eye sobriquets), the Torygraph one is just demagoguery, the Grauniad piece is better thought through.

Corbyn diluted the message for political reasons, but it’s an important discussion.

First out of the post was South Korea, which is currently ruled by the liberal Min Ju party. In early August, the Koreans announced that tax incentives would be limited on automation investments.

This is not a ‘robot tax’ as such, but it does recognize that if the state provides a safety net for its citizens, that service must be funded by society.

Traditionally, this has been paid for by corporations and job-holding citizens, and a strong shift toward automation means that more citizens will lose their jobs—if we assume for this analysis that demography remains unchanged, then governments will find it increasingly difficult to support their citizens.

The choices are stark, but partial options could be combined.

  • The Luddite option—freeze automation
  • The Robot Tax—increase revenue from companies which reduce their workforce
  • Increase debt—business as usual, pretend the problem doesn’t exist
  • Reduce benefits—when a threshold is broken, there will be blood on the streets

The alternative view to all this is that Artificial Intelligence (AI) will create more jobs than it destroys. That’s one area where the debate is particularly hot.

The three buses problem. Transport researchers have built mathematical models to study this problem (hint: it never happens on the underground).

PwC put out a press release on AI and jobs in March 2017, which is disturbing on two counts—the numbers are compelling, but the interpretation is weak.

The study states that up to 30% of UK jobs will be gone by 2030, but ‘this should be offset by job gains elsewhere in the economy.’

The only suggestions for that last part are that a higher level of education will be needed for those new jobs, and they will be more social in nature.

PwC also tells us that in the US, the job loss number is 38%, and in Germany, 35%.

So there’s one key question—which side is right: AI job gain or AI job loss?

To find out, I asked a machine.

“Google, what new jobs will be created by artificial intelligence?”

A study by Accenture helped me out. Apparently, there are three fascinating entirely new job categories. These are:

  • Trainers
  • Explainers
  • Sustainers

I’ve abridged some of the explanatory text below, because in humans, tedium can easily set in.

Humans in these roles will complement the tasks performed by cognitive technology, ensuring that the work of machines is both effective and responsible.

Trainers

This first category of new jobs will need human workers to teach AI systems how they should perform…
…they teach AI algorithms how to mimic human behaviors.

Customer service chatbots, for example, need to be trained to detect the complexities and subtleties of human communication…
…Yahoo engineers have developed an algorithm that can detect sarcasm on social media and websites with an accuracy of at least 80%.

Consider, then, the job of “empathy trainer” — individuals who will teach AI systems to show compassion…
…Humans are now training the Koko algorithm to respond more empathetically to people who, for example, are frustrated that their luggage has been lost, that a product they’ve bought is defective, or that their cable service keeps going on the blink even after repeated attempts to fix it.

Without an empathy trainer, Alexa might respond to a user’s anxieties with canned, repetitive responses such as “I’m sorry to hear that” or “Sometimes talking to a friend can help.”

The second category of new jobs — explainers — will bridge the gap between technologists and business leaders. Explainers will help provide clarity, which is becoming all the more important as AI systems’ opaqueness increases. Many executives are uneasy with the “black box” nature of sophisticated machine-learning algorithms, especially when the systems they power recommend actions that go against the grain of conventional wisdom.

I can think of a couple more categories ending in ‘ainer’ for the guys who wrote the study. I would also say that all these amazing jobs are centered on humans helping machines, not machines helping humans—maybe the report was written by a robot.

Enter Eric Schmidt, your man from Google. Speaking at the Viva Tech conference in Paris in June this year, Schmidt quoted a McKinsey study that states 90% of jobs are not fully automatable.

Two points come to mind: the first is that if 90% are not, 10% are—add that to the present jobless rate. The second is the definition of fully. If we think very conservatively, and speculate that fully means only 20% (i.e. you still do the other 80% of your job, presumably for 80% of the pay), then the added employment loss is a further 18%.

Of course, you might do 100% of what you did before in 80% of the time, because AI is helping you out.

For instance, let’s say you have a job processing expense claims. When you get to work, you say good morning to your three colleagues and sit down at your desk. There’s a stack of paper invoices in front of you.

AI now provides a machine where you dump the lot, sort of like a juicer.

The machine sorts through everything, regardless of size, scans and reads issuers, dates, and amounts, and produces a spreadsheet with the results. It compares that with a sheet you’ve received from the claimant, and attempts a match. It flags any inconsistencies.

Your job is to run through the line items, query any expense that seems unjustified, or any amount entered that doesn’t match. A job that took one hour is done in fifteen minutes, so you can now process four such claims per hour—congratulations, your productivity just quadrupled.

But wait… for this to work, you need four claims on your desk every hour, and the limiting factors are: (i) how many claims you actually get; (ii) whether the speed with which your department processes them (pre-AI) introduces delays.

If your team is working well, then with the introduction of AI it now has four times the productivity, but unfortunately, not four times the work, because expense claims are not going to quadruple.

Your company is pleased as punch. You’re their star operator. It fires your three colleagues, and the departmental productivity quadruples. Actually, now it even goes up a little more—because you have no one to chat with, you can now manage a claim every twelve minutes, so you’re doing five times better.

Your new robotic colleague always says: Hi! I’m done with this batch, please feed in the next documents. It doesn’t know about your lunch hour, so it repeats this mantra at regular intervals when you’re munching your sandwich. Since it gets no input, the pitch of the automated voice shifts from cocktail lounge seductive to low-cost airline lounge wife.

Over the last few months, the damn thing has been driving you nuts. This afternoon, you weren’t quite yourself, and smarty-pants AI (you call it SPAI) said it once too often.

You hurled it out the window, two floors down—it landed on top of a parking robot and shattered its triangulation vision unit. SPAI’s last croak was “Hi! I’m done…”

Your section head wanted to keep you on—anyone can make a mistake, it’s known as human error. Unfortunately, Health & Safety had the last word. After all, if the claims robot had killed a human, can you imagine the publicity?

ROBOT CLAIMS ITS LAST VICTIM! WHAT’S THE COST FOR THE HUMAN RACE?

So there we are—all four jobs gone, but the good news is the new machine is far more advanced, and benefits from a cutting edge AI training algorithm, so it doesn’t need a human at all. And when it’s done with this batch, it turns itself off until the next one arrives.

That’s excellent for carbon emissions, and the new spy never says a word.

The India Road, Atmos Fear, Clear Eyes, and Folk Tales For Future Dreamers. QR links for smartphones and tablets.

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: